Belfius CEO Olivier Onclin on AI, trust, and sovereignty

Your banking app has already completely transformed everyday banking. The next shift’s even bigger—banking that feels like a conversation. At the end of April, just a few days before becoming CEO of Belfius, we engaged with Olivier Onclin to unpack what it truly takes to move from traditional digital channels to conversational AI, and why the "wow" factor only matters if the underlying processes are rebuilt end to end.

We get practical about how AI in banking has evolved over decades, from scripted automated decision making in credit models to compliance automation, and now to large language models that can reshape the client experience. Olivier explains why efficiency and customer experience are tightly linked, why the talent challenge is real even for well-prepared organisations, and why late starters still need to jump in quickly while treating AI as a management journey, not a plug-in tool.

Next, we explore trust in AI. Olivier breaks it down into four clear dimensions: transparency (people should know they're talking to a machine), security (including cybersecurity for consumers), a human layer (AI cannot replace basic humanity), and sovereignty. When it comes to sovereignty, we discuss European AI and why Belfius favours European providers when performance is comparable, including a partnership with Mistral, alongside the broader need for resilient infrastructure and cooperation across European players.

We also tackle the uncomfortable topic of ROI. If the best transformations can take seven to ten years to pay back, how do leaders justify AI investments in uncertain times and short reporting cycles? If you care about digital transformation, AI governance, cybersecurity, and the future of customer experience in financial services, this conversation will provide a sharp framework and real-world constraints. 

If this episode has sparked new ideas on conversational banking, AI governance, and responsible AI in your organisation, subscribe, share it, and leave a review.

Join us and listen to all episodes on www.pwc.be/aiunscripted

Transcript

Welcome and introduction of Olivier Onclin

SPEAKER_01 0:00

AI unscripted. This is a new episode, and as you have already seen in previous episodes, this is a podcast talking about AI, of course, but with a bit of a touch with focusing on AI from a business perspective. And today again I'm not alone. I'm here with uh Olivier Enclin, CEO of Belfius. Welcome, thanks for joining me.

SPEAKER_02 0:23

Thank you.

SPEAKER_01 0:24

You're new, relatively new in your role. Um so maybe a bit of context who you are, uh what you've done before.

SPEAKER_00 0:40

Think of me as a banking drummer or a drumming banker because uh my early career choice had to be made between music and banking. I made the banking choice in the end. I will leave it up to the others to decide whether that was a good choice. Yeah. Um I had a career of 10 years at Euroclear, which I think the greater public knows by now, thanks to the Russia episode. And after 10 years, I um actually joined the DEXA group at the time. Uh was uh actually witnessing the financial crisis from very close and then uh helped build uh Belfast to what it to what it became today, and uh very proud with a team that uh we can share this successful story so far.

SPEAKER_01 1:27

Yeah, and how are your CEO?

SPEAKER_00 1:29

I uh will be CEO officially uh in a in a couple of days. Yeah, yeah, very privileged.

SPEAKER_01 1:36

Of course, in an era we are talking specifically on on AI, but also geopolitically, macroeconomic, it's not an easy time of a lot of uncertainty.

Podcast discussion

SPEAKER_01 1:45

So specifically focus on NAI. What is the group, what is Bapheus doing with respect to AI, in fact?

SPEAKER_00 1:52

That's uh that's a question with a very long answer because uh of course AI has come to the attention as now large language modeling. Um, but we have been using AI since many years, I would say even since decades, since we introduced like automated decision making, which was heavily scripted in our credit decisioning models that we have been doing for a long time now. Um, the last years we had a strong focus on making compliance uh very much more efficient because the regulation was getting so uh intrusive uh with AI, and now we are in a phase where we are reinventing, I think, uh the digital interaction model with our clients thanks to AI. Yeah. Um and I think that that transition will be probably long, but it will be very impactful. Um if I if I make the parallel between in 2012, we made a decision of heavily investing in a mobile application because we were convinced that the digital interaction would heavily focus on a mobile interaction channel. Uh, that is absolutely the case and is is standard, I would say, across the industry today. Well, tomorrow that interaction model is going to change to fully conversational. And that cannot happen without AI.

SPEAKER_01 3:13

And that's basically the client experience. So because I think there are two aspects um typically, of course, also in your case, I guess, the efficiency gain and then basically the user experience. The older one, well, the beginning was maybe more an efficiency gain. Is that is that fair to say that it started with gaining efficiency and that now in the last years it's more the user experience, is or is it not that binary?

SPEAKER_00 3:38

No, I would say the two are very strongly connected because if you really want to have a native digital experience for your client, it needs to be an end-to-end implementation which actually automatically leads to the efficiency gain as well, or the like uh avoiding of having more workload linked to a certain process. And and that is what we have been witnessing now, uh, where you could uh have a heavily simulate a digital experience by having a very interesting front-end experience with the new conversational wave which is coming, it will need to be end-to-end implemented, and the investments related to that are much more important than what we have seen in the past.

SPEAKER_01 4:20

And does that mean that the transformation journey that many other uh sectors companies are going through also to get their own people on board uh with AI, that this is less of a challenge for Belfius?

SPEAKER_00 4:35

Well, we have been preparing for that for a long time. Exactly. Yeah, exactly. And so I think uh we have like a learning curve which has been steep and long, and we are ready, I guess, to do that deep transformation. Uh and well, to to to tackle your question differently, I think it will be a serious challenge for some other companies because that talent is not very widely spread, I would say, and even also for us, will push us at a certain point to question whether we should also look for talent in a broader region than just uh the Belgian uh national territory.

SPEAKER_01 5:15

And for the ones that haven't started it well years ago, as you have done, but what would then the lesson learned be for them? Because of course now it's already a completely different landscape. Uh, the speed of change is that fast that the train has already left to some extent the station. So, how do they jump on that train still?

SPEAKER_00 5:35

I I think the only positive thing maybe is that technology has known such an evolution that you can reap the benefits of the technology which is at hand today. Um, but uh technology in itself is not a solution, of course. Uh it it is a whole way of designing things and thinking about things that has to evolve together with it, and that takes some time and takes a lot of uh management attention and management involvement. So it is a journey, however you turn it. And and so I I would say get started and make sure everybody's involved.

SPEAKER_01 6:12

And maybe talking about trust, because you're of course in a sector highly regulated, but also trusted everything. And AI is not always the right uh avenue towards the the truth or the outcome. How does that work, in fact?

SPEAKER_00 6:28

Um we we have uh been giving that a lot of thought. And I would say I would break it down in in four dimensions trust. First of all, when you use AI, I think you need to be transparent about it. That is very important. People need to know that they are interacting with a machine and not with a human. That is absolutely number one. Two, AI needs to come with a certain, I would say, a big degree of security. And of course, I'm talking about cybersecurity for us as a company, but I don't think the major problem for the future lies with banks which are heavily regulated and which have been preparing for major cybersecurity risk for quite some time now. We also need to think about the cybersecurity of the end consumer. And we see that this is becoming a societal problem, and we need to think about it how can we get the society as resilient as possible in this new era of AI. And that involves uh not only banks, that involves telco players, that involves the end consumer as well, that needs to have a higher degree of understanding of what's going on in order to become uh more resilient. Um, another dimension I would say is that I don't see AI without any human link to it. What do I mean with that is that it's it's our strong belief that even if we think we can displace a lot of activity into a digital context with an AI-powered engine behind it, there needs to be a moment that a consumer can ask to go to a human intervention. And and and I think that is also not in the end game.

SPEAKER_01 8:12

When basically in two, three years' time that you still think will be it's still gonna be at the end human.

SPEAKER_00 8:20

Well, I think that the role of the human is fundamentally gonna shift. But I think there will still be an human intervention needed, but that the thing that the human will bring will be again to some kind of basic humanity relationship, which AI will not bring. The last dimension in security might sound strange or not. I don't know in these times, it's sovereignty for me. I think we need to think, and especially the corporate world, about more independence in terms of this technology. This technology is becoming core to our functioning, is becoming core to our societal functioning. And so we need more European solutions also to be able to implement them. And so at Belfast, what we have chosen is when we have solutions that are at par from European providers, we will privilege European providers because it helps pushing the European industry forward in that industry that is becoming fundamental to our sovereignty.

SPEAKER_01 9:22

And you have sufficient choice to buy Europe?

SPEAKER_00 9:26

Well, if we if we're talking about um large language models, uh there is one basically European solution, it's Mistrel, and we have uh concluded a partnership with them. Okay. Uh I cannot say we are exclusive with Mistrel, but everywhere, and we keep on testing that, that the model is at par with other models, we privilege them. That's how we we tackle it for the moment. But it's not only about large language models. There's a basic infrastructure which needs to be reviewed in terms of data centers. I think there is a lot of thought and work to be done also in terms of cooperation between European players, of making that infrastructure more localized and more on European territory. Because it it's a bit like the defense industry. It's that critical to the functioning of society that I think we need to go into that.

SPEAKER_01 10:10

Before you know it, you have replaced one dependency by another, and AI could be another one in the case.

unknown 10:15

Yeah.

SPEAKER_01 10:16

Maybe talking about um uh a slightly different uh topic, and and that's of course when we talked about in the past bitcoins and and so on, we talked about the role of banks as the middleman. Is that going to have an impact, yes or no? The question is whether AI will also not put another layer on that question. And I asked the question because as a consultant, we're also to some extent the middleman between data and sometimes governments, where you can ask questions is that going to be a role going forward, or can they just cut out the middleman? So I do you think that that in terms of the basic um role of a financial institution, that AI will again put an additional layer of of complexity on that, on on top of crypto, and that we may uh maybe a difficult question, but well, i if I look at it uh a little bit philosophically, I I would say that AI is kind of uh making uh a democratic access to data possible for everyone.

SPEAKER_00 11:20

Yeah. So if if you think about that, then you would say in general, the tendency will be that consumers are get more informed and better prepared. So I more see it in uh an industry as ours, which is actually an advice industry, a human advice industry, that it will probably shift the role very much of what the human uh play will be in our industry. But as I said before, I'm a strong optimist in the role of the human in our industry. We have seen that I'm gonna make a strange parallel maybe but with private banking. We started a private banking strategy uh in 2019, and we decided to like develop a new physical distribution network, which was totally against the current. All banks were scaling down their physical networks. And what do we see where everybody was predicting that the physical networks would disappear? We strongly increased it, and there is a perfect linear correlation between our performance where we have this physical distribution point and where we don't have them. And so there you see that there there is this role. People basically want to connect with other humans, especially another selective number of topics that it will displace a lot of work, but it will not replace the human.

SPEAKER_01 12:43

Yeah. Which of course that means well triggers the question, um, also as a society as a whole, as a country as a whole, if you're gonna be able to bring everyone to that new reality, of course. That's of course also a philosophical question, but an important one also.

SPEAKER_00 13:00

Absolutely. And we have a tendency to like over over expect with regards to the speed at which it's gonna happen. Yeah. It's the same with digital in essence. And and so I think it will happen eventually, but it's gonna take some time. And we also need to make sure, and I think that's a kind of a societal responsibility, that we leave no one behind.

SPEAKER_01 13:19

Yeah. And how do you see that within Bellfields in terms of upskilling the massive amount of learning and development, I guess, that you also need to have a lot of people.

SPEAKER_00 13:28

Yeah, we have, of course, for our own staff, uh, we have learning and development. But we also see uh and and research clearly shows that that financial literacy in itself in Belgium is really not going so well. And there we see a role also for digital platforms such as ours, which have a lot of reach in society, that we need to play some kind of a role of making this dig uh this financial literacy more um yeah, uh like a process that that that that invites people to to learn more about finance, and we are heavily thinking on how to to make that a reality.

SPEAKER_01 14:03

Which is an interesting one because in previous sessions that we had in the podcast, we also talked to to, for example, the VUB, where indeed uh then it wasn't the financial literacy, but it's basically also in terms of literacy and what AI really means. Because sometimes we are thinking it's a bit Google type of search uh on steroids, but I think it's a bit more, of course. So that's a very interesting uh topic. So how how do you then see a f uh uh a group like Belfius? Yeah. What will be Belfius in three, four years' time? How will you look like then?

SPEAKER_00 14:39

Well, uh in three to four years' time I tend uh to think that we overestimate the impact. Okay, which does not mean that we don't need to worry about this today. Because as we said, it is a journey, and you need to transform a whole uh company in order to prepare for that journey. So so I think AI will become uh naturally integrated in our day-to-day way of working, in our solutions, in our uh decision mechanisms going forward. But I don't think we will see in three to four years a company which is fundamentally different from what it is today. Okay, so an optimist, that's basically you have to be an optimist. I mean it it's it's a kind of a duty to be optimistic, and that's not something naive, but that's believing that we can create progress.

SPEAKER_01 15:32

Yeah, yeah, yeah. And how do how optimistic are you then of the economic environment in a region like Belgium, but also Europe? Because of course a lot of you talked about being a bit more independent technology-wise. Of course, that's also the reality uh that a lot of players are not European. Do you see that we have the platform in Europe to to leverage on that? A bit less capital intense, but more brain.

SPEAKER_00 15:59

But you you know, uh that's a question that that actually uh occupies me a lot. Um, because what I find um strange, so to say, is that people have a tendency to criticize a lot European politics. Yeah. And I do have to agree on one thing, we could do much better. Yeah. But at the same time, there is some kind of a tendency in all countries to push responsibilities down to lower levels. And so there is some kind of a well, uh, I will I will get there. Uh uh, between you don't want to push responsibilities to a higher level of European sovereignty, but you criticize Europe for not being able to do or to take the decisions it needs to take. And so I think we should stop like shifting all responsibilities to the political level. I think as a corporate, there's a lot of things we can do. We can instigate more cooperation between European companies. And I must say, I'm again there optimistic because because of what's happening in the geopolitical world, I see discussions now that we can hold with other European banks about investments in uh mutualized platforms which were not possible a couple of years ago. So there I think we will see a shift in cooperation and the creation of some kind of European ecosystem.

SPEAKER_01 17:25

But I think you're right, because you are in fact we were saying that we're always saying that someone else is the culprit, is the one that doesn't, but it's also looking at ourselves, that's a bit the point. But of course, there's also the the the fact that in terms of yeah, the integrated capital markets, that's also part of your sector, of course, but that's of course not that developed in Europe than it is in the US. And then you see if you need one, two capital rounds, well, the people that we have, brainy people, they go, they leave. So will you do you see that there we would be able to take steps uh forward?

SPEAKER_00 17:59

Because we need a we would need a unified capital market, absolutely. But is that the silver bullet solution? I don't think so. There needs to be a cultural shift as well. Because if you if you look, for instance, a statistic that always like kind of strikes me is that for every 100 euros invested by a European household, there's 65 euros invested in the US. We should ask ourselves questions about that. What are we doing? We are financing growth somewhere else. We are pushing our our our champions to go to go somewhere else. And so I don't want to preach protectionism, but we need to be conscious about that. And we need to be more European conscious if we really want to m instigate some kind of a change here.

SPEAKER_01 18:43

Yeah. Because we also see about the NATO spend. Uh one percent NATO spend for us is one percent, almost one percent uh real spent because it doesn't flow back. That's the similar philosophy. Yeah, yeah. If you were basically 18 again, imagine. What would you study now? I don't know what you have studied.

SPEAKER_00 19:05

I studied the commercial engineering. Okay.

SPEAKER_01 19:07

Would you do it again?

SPEAKER_00 19:09

I don't think so. To be to be honest. Well, the thing is, I mean, it's it's interesting because you kind of get to know yourself along the way and get to know your strengths along the way. And so I probably would have done something completely different. Now, again, I'm very happy I studied that because otherwise I wouldn't have not been sitting here. Right. But but I think I would have gone for something more creative as a study.

SPEAKER_01 19:34

Yeah, you talked about drumming.

SPEAKER_00 19:36

Yeah, but like architecture. I'm I'm absolutely uh very interested and passionate about architecture, so I think I would have studied that. But I come from a time, and uh this might uh sound like a very old man, but where you could be a doctor, you studied economics or law. That were basically the three choices. We we were not really aware of all the rest that existed, at least not me.

SPEAKER_01 19:57

Yeah, and of course it's for us difficult to project your back. Yeah. I think the current environment, of course, is is yeah, we are not AI natives or whatever else. To say the least, yeah.

Rapid Fire segment

SPEAKER_01 20:11

Maybe to um to land a bit this conversation uh with a few very short, snappy uh questions, don't think too long about the the answers. You can, but it doesn't matter. Um the first thing that you do with AI on a day, what is that?

SPEAKER_00 20:28

Well, um my daughter uh asked for a Chat GPT uh subscription, paying subscription before I was really very aware of what's good what was going on. And since then I share that subscription with her, and I go for a conversation every day in OpenAI just to keep myself a bit aware of what it can and what it can't do. And if you challenge it, it's pretty interesting to see what uh what can happen.

SPEAKER_01 20:53

I know what you will answer, but it doesn't matter. Studying in Belgium or studying in the US?

SPEAKER_00 20:58

No, in Belgium.

SPEAKER_01 21:00

Okay. Mastering in the US or mastering in Belgium?

SPEAKER_00 21:03

Mastering in Europe. In Europe, okay.

SPEAKER_01 21:06

What's the best alternative for Belgium in Europe then?

SPEAKER_00 21:09

Oh, I don't think there's one single best alternative. Um, there's uh many alternatives I would consider. I mean, uh I have a natural link to Scandinavia because I'm married to Scandinavian. Um, but uh there's also schools and uh yeah programs in the south of the Europe which are pretty impressive.

SPEAKER_01 21:27

Okay bringing a sense of urgency to the people that indeed maybe we are overestimating stuff today, but it will come in terms of uh impact. Fear or opportunity, what's the best thing to sell that argument?

SPEAKER_00 21:46

I think it's definitely opportunity, yeah. But it needs to be really like in it to also our educational system. And I don't think our educational system today is ready. To make that opportunity a possibility.

SPEAKER_01 22:03

And you already touched upon the fact that as economic players we also have a role to play. But do you see changes happening there and are they going fast enough? And do you need to start with the primary schools or do you need to start with universities your way back? How how do you basically where do you start? Because it's an incredible complex uh problem, of course.

SPEAKER_00 22:28

I would start as soon as possible. Because otherwise anyway we lose control of it. I think today we are in a situation where um youngsters find their own way on how to interact uh with the digital world before they get some kind of a context or guidance uh in in their schooling or educational system, and I think that should change.

SPEAKER_01 22:49

Okay. And then I have the last question, which is uh probably again a topic for an entire session, but it is people don't see the return on investment. Uh that's also when we look a bit at uh all kinds of barometers and surveys that we do, various industries, people are a bit um underwhelmed about uh not the investment that they need to do because they are overwhelmed, but underwhelmed about the impact, the effect on their PL. Um how do we turn that around? Um that people want instant uh satisfaction almost and say, okay, I spend a new euro on AI. Um I want to have a euro saving or a euro extra growth. It's not that easy, of course. But how do you how do you bring that across?

SPEAKER_00 23:37

Also as management, because yes, you're investing in difficult times, money, and the return is not necessarily No, and and and it's it's it's a very, very good question because if we look at our most successful developments, these developments have typically actually paid back over a period of I would say even seven to ten years. I I'm gonna take the example of our bank insurance model. We have heavily invested in our insurance model. Well, that took 10 years. If you make a business case, you never make a business case on 10 years. If you have a mandate as a CEO or management team, you never have a mandate of 10 years. And so we are coming in a complex world where we have to make decisions which are far beyond actually our management time or our management term. And and I think that makes uh the world to where it is today where where where we see this disruption that is possible by new players, because incumbents are not in a position where they make the the right decisions. But I I think we have been uh very uh well very happy with our shareholder, uh, which has had until now that long-term view and has put us in a place where we could make the investments which we were we deem necessary in order to have long-term value creation. But I I do realize that it's not uh so easy uh if you're chasing every three master's results.

SPEAKER_01 25:03

Exactly, because I see more and more groups that are basically saying, yeah, there you need to have an a return on investment almost in the year euro for euro, but that it doesn't work like that.

SPEAKER_00 25:13

It doesn't work, and for us, it's certainly not the main KPI we look at.

SPEAKER_01 25:18

Okay, perfect.

Closing of the conversation

SPEAKER_01 25:20

I will leave it here. Thank you very much for uh the insights. You're very welcome. Here and there a bit provocative, that's fine. I like it. So, and yes, uh, from my perspective, thanks for tuning in, and yeah, hope to see you uh soon again on another episode of AI on scripted. Bye bye.